Prediction with Time-Series Mixer for the S&P500 Index

Junyi Ye, Jingyi Gu, Ankan Dash, Fadi P. Deek, Guiling Grace Wang

Research output: Chapter in Book/Report/Conference proceedingConference contribution


As an essential US economic indicator, the S&P500 Index is used to assess the current state of market performance and gauge the economy's future course. However, stock market index prediction is challenging due to its nonlinearity and inherently volatile character. Recurrent Neural Networks (RNN) and their variants are de facto standards for sequence modeling. Recently, Convolutional Neural Networks (CNN) and attention-based networks, such as dilated casual convolutions and Transformers, have also become popular in time series forecasting. In this paper, we report on the design of a Time-Series Mixer (TS-Mixer) architecture based on MLP-Mixer, an all-MLP architecture for time series forecasting. To the best of our knowledge, this is the first implementation of MLP-Mixer-based architecture for sequence modeling. Modern deep learning models are increasingly built to handle univariate time series data. They generally pay attention to analyzing temporal dependencies while ignoring the relationship among features. The proposed architecture is specifically created for multivariate time series forecasting to capture temporal feature interactions while simultaneously learning feature correlations. To accomplish this, the proposed Time-Feature Mixer contains two types of MLP layers: feature mixer and temporal mixer. The feature mixer is applied independently to each data point to capture the correlation among features. In contrast, the temporal mixer extracts temporal dependency (trend, seasonal, cyclical, or random characteristics) of each feature across the whole input sequence. Compared to prevalent neural networks in sequence modeling, TS-Mixer exhibits competitive performance regarding S&P500 Index prediction.

Original languageEnglish (US)
Title of host publicationProceedings - 2023 IEEE 39th International Conference on Data Engineering Workshops, ICDEW 2023
PublisherInstitute of Electrical and Electronics Engineers Inc.
Number of pages8
ISBN (Electronic)9798350322446
StatePublished - 2023
Event39th IEEE International Conference on Data Engineering Workshops, ICDEW 2023 - Anaheim, United States
Duration: Apr 3 2023Apr 7 2023

Publication series

NameProceedings - 2023 IEEE 39th International Conference on Data Engineering Workshops, ICDEW 2023


Conference39th IEEE International Conference on Data Engineering Workshops, ICDEW 2023
Country/TerritoryUnited States

All Science Journal Classification (ASJC) codes

  • Information Systems and Management
  • Statistics, Probability and Uncertainty
  • Management Information Systems
  • Artificial Intelligence
  • Computer Networks and Communications
  • Information Systems
  • Decision Sciences (miscellaneous)


  • MLP-Mixer
  • multi-layer perceptron
  • neural networks
  • stock market prediction
  • time series forecasting


Dive into the research topics of 'Prediction with Time-Series Mixer for the S&P500 Index'. Together they form a unique fingerprint.

Cite this